Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
17
Linear and Loop Antennas
The Hertzian dipole, uniform line element, and small linear dipole examples do not
satisfy Eq. (17.1.2), except when the antenna length is electrically short, that is, l .
For loop antennas, we may take the loop to lie on the xy-plane and be centered at the
origin. Again, we may assume a thin wire. For a circular loop of radius a, the current
ows azimuthally. The corresponding current density can be expressed in cylindrical
coordinates r = (, , z) as:
I( a)(z)
J(r)=
(circular loop)
(17.1.3)
The delta functions conne the current on the = a circle on the xy-plane. We will
discuss loop antennas in Sec. 17.8.
Antenna arrays may be formed by considering a group of antenna elements, such as
Hertzian or half-wave dipoles, arranged in particular geometrical congurations, such
as along a particular direction. Some examples of antenna arrays that are made up from
identical antenna elements are as follows:
J(r) =
z
an I(z)(x xn )(y)
an I(z)(y yn )(x)
an I(z zn )(x)(y)
2D planar array
J(r) =
z
n
z
J(r) =
n
z
J(r) =
J(r)=
z I(z)(x)(y)
(17.1.1)
mn
The weights an , amn are chosen appropriately to achieve desired directivity properties for the array. We discuss arrays in Sec. 20.1.
It is evident now from Eq. (17.1.1) that the radiation vector F will have only a zcomponent. Indeed, we have from the denition Eq. (15.7.5):
where I(z) is the current distribution along the antenna element. It is shown in Sec. 22.4
that I(z) satises approximately the Helmholtz equation along the antenna:
d2 I(z)
+ k2 I(z)= 0
dz2
(17.1.2)
F=
Hertzian dipole
Uniform line element
Small linear dipole
Standing-wave antenna
Half-wave antenna (l = /2)
Traveling-wave antenna
where l is the length of the antenna element and the expressions are assumed to be valid
for l/2 z l/2, so that the antenna element straddles the xy-plane.
J(r )ej kr d3 r =
z
The x and y integrations are done trivially, whereas the z integration extends over
the length l of the antenna. Thus,
l/2
I(z)= Il(z)
I(z)= I
I(z)= I(1 2|z|/l)
I(z)= I sin k(l/2 |z|)
I(z)= I cos(kz)
I(z)= Iejkz
F=
z Fz =
z
l/2
Using Eq. (15.8.3), the wave vector k can be resolved in cartesian components as:
k cos sin + y
k sin sin +
kx + y
ky +
k = k
r=x
zk cos = x
z kz
Thus,
kx = k cos sin
ky = k sin sin
kz = k cos
(17.1.4)
777
Fz ()=
l/2
jkz z
I(z )e
l/2
dz =
l/2
jkz cos
I(z )e
dz
It follows that the radiation vector Fz will only depend on the polar angle :
l/2
778
Umax =
(17.1.5)
k2
|Il|2
322
g()=
H =
jk
H=
45
0o
45o
45
45
(17.1.6)
ejkr
Fz ()sin
4r
0.5
90
135o
135o
90o
90
6 3
dB
90o
135o
135o
180o
k2
|Fz ()|2 sin2
322
(17.2.1)
Gain in dB
The elds are omnidirectional, that is, independent of the azimuthal angle . The
factor sin arises from the cartesian to spherical coordinate transformation, whereas
the factor Fz () incorporates the dependence on the assumed current distribution I(z).
The radiation intensity U(, ) has -dependence only and is given by Eq. (16.1.4):
U()=
The gain g() is plotted in absolute and dB units in Fig. 17.2.1. Note that the 3-dB
or half-power circle intersects the gain curve at 45o angles. Therefore, the half-power
beam width (HPBW) will be 90o not a very narrow beam. We note also that there is no
radiated power along the direction of the antenna element, that is, the z-direction, or
= 0.
jkr
E =
jk e
Fz ()sin
E=
4r
U()
= sin2
Umax
180o
(17.1.7)
Fig. 17.2.1 Gain of Hertzian dipole in absolute and dB units.
To summarize, the radiated elds, the total radiated power, and the angular distribution of radiation from a linear antenna are completely determined by the quantity
Fz () dened in Eq. (17.1.5).
In these plots, the gain was computed by the function dipole and plotted with abp
and dbp. For example the left gure was generated by:
[g, th, c] = dipole(0, 200);
abp(th, g, 45);
l/2
Fz ()=
l/2
l/2
l/2
Il(z )ejkz
cos
g() sin dd = 2
g() sin d = 2
dz = Il
k2
U()=
|Il|2 sin2
322
2
=
8
3
Dmax =
4
4
= 1.5 1.76 dB
=
8/3
sin3 d ,
0
or,
779
8
k2
k2 |Il|2
|Il|2
=
2
32
3
12
h
(17.2.2)
Because of the proportionality to |I|2 , we are led to dene the radiation resistance
of the antenna, Rrad , as the resistance that would dissipate the same amount of power
as the power radiated, that is, we dene it through:
Prad =
1
Rrad |I|2
2
(17.2.3)
2
k2 l2
=
6
3
2
l
Fz ()=
2I cos(kh cos ) cos(kh)
I sin k(l/2 |z |) ejkz cos dz =
k
sin2
Inserting Fz () into Eq. (17.1.7), and canceling some common factors, we obtain:
U()=
(17.2.4)
du(z)
u(z + l/2)u(z l/2)
Il
= Il(z)
I(z)= Il
l
dz
It follows that the normalized power gain g() will have a similar form:
I(z)= Ik
l
2
|z| ,
l
2
(17.3.3)
l
2
(17.3.1)
(17.3.4)
The radiation resistance may be dened either in terms of the peak current or in
terms of the input current through the denitions:
1
1
Rpeak |I|2 = Rin |Iin |2
2
2
Rin =
Rpeak
sin2 kh
(17.3.5)
When l is a half-multiple of , the input and peak currents are equal and the two denitions of the radiation resistance are the same. But when l is a multiple of , Eq. (17.3.4)
gives zero for the input current, which would imply an innite input resistance Rin . In
practice, the current distribution is only approximately sinusoidal and the input current
is not exactly zero.
The input impedance of an antenna has in general both a resistive part Rin and a
reactive part Xin , so that Zin = Rin + jXin . The relevant theory is discussed in Sec. 23.3.
Assuming a sinusoidal current, Zin can be computed by Eq. (23.3.10), implemented by
the MATLAB function imped:
Zin = imped(l,a);
(standing-wave antenna)
(normalized gain)
Prad =
and we must assume, of course, that the product Il remains nite in that limit.
I(z)= I sin k(l/2 |z|)
(17.3.2)
For a general dipole of length l, the current at the input terminals of the antenna is
not necessarily equal to the peak amplitude I. Indeed, setting z = 0 in (17.3.1) we have:
The Hertzian dipole may be thought of as the limiting case of this example in the limit
l 0. Indeed, multiplying and dividing by l, and using the property that the derivative
of the unit-step is u (z)= (z), we have
2
|I|2
cos(kh cos ) cos(kh)
2
8
sin
cos(kh cos ) cos(kh) 2
g()= cn
sin
Rrad =
Prad = Umax =
780
where l, a are the length and radius of the antenna in units of . For example, a half-wave
dipole (l = /2) with zero radius has Zin = imped(0.5, 0)= 73.1 + j 42.5 .
For l
a, the input resistance remains largely independent of the radius a. The
reactance has a stronger dependence on a. Fig. 17.3.1 shows a plot of Rin and Xin versus
781
Reactance
Resistance
250
782
the above integral can be expressed as a sum of two integrals of the form:
800
600
200
150
X, ohm
R, ohm
400
100
200
200
400
a=0
a = 0.0005
a = 0.005
50
600
0
0.3
0.4
0.5
0.6
l/
0.7
800
0.3
0.4
0.5
l/
0.6
2
Cin (kl)+
0.7
the antenna length l plotted over the interval 0.3 l 0.7, for the three choices of
the radius: a = 0, a = 0.0005, and a = 0.005.
We observe that the reactance Xin vanishes for lengths that are a little shorter than
l = /2. Such antennas are called resonant antennas in analogy with a resonant RLC
circuit whose input impedance Z = R + j(L 1/C) has a vanishing reactance at its
l = 0.4857,
l = 0.4801,
l = 0.4681,
Rin = 67.2
Rin = 65.0
Rin = 60.5
An analytical expression for the peak and input radiation resistances can be obtained
by integrating the radiation intensity (17.3.2) over all solid angles to get the total radiated
power:
Prad =
=
2
U() d =
2
|I|
4
0
U()sin d d = 2
U()sin d
0
2
d
Rpeak =
0
2
1
cos kl 2Cin (kl)Cin (2kl) + sin kl Si (2kl)2Si (kl)
2
2
2
1
cos(2kh cos ) cos(2kh) 2 cos(kh cos ) cos(kh) cos kh
2
(17.3.6)
1
Cin (kl)+ cos kl 2Cin (kl)Cin (2kl) + sin kl Si (2kl)2Si (kl)
2
2
2
(17.3.7)
which agrees with Eq. (23.3.21) derived by a different method. The radiation resistance
Rpeak also determines the directivity of the dipole antenna. Using (17.3.3) for the normalized gain, we nd the beam solid angle:
2
=
g() d = 2cn
2
d = 2cn
2Rpeak
Dmax =
4
1
cn Rpeak
(17.3.8)
The normalization constant cn is equal to unity for a half-wave dipole; for other
antenna lengths, it may be computed numerically.
The MATLAB function dipdir calculates cn , the directivity Dmax , the angle max at
which the directivity is maximum (the angle 180 max also corresponds to Dmax ), and
the radiation resistance Rpeak . It has usage:
[Rpeak,Dmax,thmax,cn] = dipdir(L)
The radiation resistance is computed from Eq. (17.3.7) with the help of the sine and
cosine integral functions Si (x) and Cin (x), and Dmax is computed from (17.3.8).
The table below shows some representative values, with the corresponding angular
patterns shown in Fig. 17.4.2.
l/
d =
Rpeak =
a = 0,
a = 0.0005,
a = 0.005,
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
Rpeak ()
73.08
185.68
198.95
106.46
105.42
229.94
259.45
143.48
120.68
Dmax
1.64
1.88
2.41
3.28
2.23
2.37
2.53
3.07
3.06
Dmax (dB)
2.15
2.75
3.82
5.16
3.48
3.75
4.03
4.87
4.86
max
cn
90.00o
90.00o
90.00o
90.00o
42.57o
50.94o
57.42o
62.28o
32.22o
1.0000
0.3431
0.2500
0.3431
0.5109
0.2200
0.1828
0.2723
0.3249
783
I(z)= I cos(kz)
(half-wave dipole)
(17.4.1)
784
where the beam solid angle is computed by the approximation to the integral:
g()=
= 2
(half-wave dipole gain)
sin2
(17.4.2)
Note that the maximum does occur at = /2 and the normalization constant is
cn = 1. Fig. 17.4.1 shows the gain in absolute and dB units. The 3-dB or half-power
circle intersects the gain at an angle of 3dB = 50.96o , which leads to a half-power beam
width of HPBW = 180o 23dB = 78.08o , that is, somewhat narrower than the Hertzian
dipole.
Halfwave dipole
Gain in dB
0o
0o
45o
N
1
g(i )sin i
i=0
Gauss-Legendre quadrature integration also produces accurate results. For example, assuming the normalization constant cn is known, the following code fragment
integrates the gain function (17.3.3) to compute the beam solid angle:
45o
45o
g()sin d 2
45o
90o
90o
90o
135o
135o
90o
135o
135o
180o
6 3
dB
Fig. 17.4.2 shows the gains of a variety of dipoles of different lengths. The corresponding directivities are indicated on each plot.
180o
Rin = Rpeak =
Cin (2kl)=
Dmax =
= 1.64 2.15 dB
Rpeak
In practice, the value Rin = 73 ohm can be matched easily to the characteristic
impedance of the feed line. For arbitrary values of the length l, the following example
MATLAB code used to calculate the gain function g(), as well as the constant cn and
the beam solid angle, is as follows:
Pmonopole =
1
Pdipole ,
2
Rmonopole =
1
Rdipole
2
Similarly, the directivity doubles because the isotropic radiation intensity in the denominator of Eq. (16.2.2) becomes half its dipole value:
Dmonopole = 2Ddipole
785
l = 0.50, D = 2.15 dB
l = 0.75, D = 2.75 dB
90o
6
dB
90o
6
dB
90o
45o
90o
135o
135o
45
45
90o
135o
135o
45
45
6
dB
180
180
l = 1.50, D = 3.48 dB
l = 1.75, D = 3.75 dB
6
dB
90
90
I(z)= Iejkz ,
45
45
ejkz + ejkz
45
45
l = 2.50, D = 4.86 dB
I(z)= I cos(kz)=
The backward-moving component may be eliminated by terminating the linear antenna at an appropriate matched load resistance, as shown in Fig. 17.6.1. The resulting
antenna is called a traveling-wave antenna or a Beverage antenna. The current along its
length has the form:
135
I
90
180
45
l = 2.25, D = 4.87 dB
6
dB
135
45
180
l = 2.00, D = 4.03 dB
o
90
135
135
180
45
135
6
dB
0o
45
90
135
45
45
90
45
0o
Fig. 17.5.1 Quarter-wave monopole above ground plane and the equivalent half-wave dipole.
180
0o
90o
135o
135o
l = 1.25, D = 5.16 dB
l = 1.00, D = 3.82 dB
45
786
45
0zl
(17.6.1)
90
6
dB
135o
135
180
90
90
6
dB
135o
135
180
90
90
6
dB
135o
135
l
90
F=
z
F ()= Fz ()sin =
The quarter-wave monopole antenna whose length is /4 is perhaps the most widely
used antenna. For AM transmitting antennas operating in the 300 m or 1 MHz band, the
antenna height will be large, /4 = 75 m, requiring special supporting cables.
In mobile applications in the 30 cm or 1 GHz band, the antenna length will be fairly
small, /4 = 7.5 cm. The roof of a car plays the role of the conducting plane in this
case.
We note also in Fig. 17.4.2 that the l = 1.25 = 10/8 dipole has the largest gain. It
can be used as a monopole in mobile applications requiring higher gains. Such antennas
are called 5/8-wave monopoles because their length is l/2 = 5/8.
I 1 ejkl(1cos )
jk
1 cos
(17.6.2)
180
I
1 e2jL(1cos )
I
F()
sin
jk
1 cos
jk
(17.6.3)
U()=
|I|2
|I|2
|F()|2 =
2
32
82
sin sinL(1 cos ) 2
1 cos
(17.6.4)
787
788
sin sinL(1 cos ) 2
g()= cn
1 cos
(17.6.5)
where cn is a normalization constant. Fig. 17.6.2 shows the power gains and directivities
for the cases l = 5 and l = 10, or L = 5 and L = 10.
L = 5, D = 10.7 dB, 0 = 22.2o
0o
45o
6 3
dB
90o
135
o
180
45o
90o
135o
0o
45o
45o
90o
6 3
dB
sin sinL(p cos ) 2
g()= cn
p cos
90o
For long lengths L (and for p < 1), it peaks along the direction 0 = arccos(p).
Note that p can take the values: (a) p > 1 (slow waves), as in the case of the corrugated
plane structure or the case of a Beverage antenna wrapped
in a dielectric, (b) p < 1 (fast
135o
(17.6.8)
135
180
1 2c /2 , and (c) p = 1,
The MATLAB function traveling calculates the gain (17.6.5). For example, the left
graph in Fig. 17.6.2 was generated by the MATLAB code:
The longer the length l, the more the main lobes tilt towards the traveling direction
of the antenna. The main lobes occur approximately at the polar angle (in radians) [57]:
0.371
0.371
0 = arccos 1
= arccos 1
l
L
(17.6.6)
For the two examples of Fig. 17.6.2, this expression gives for L = 5 and L = 10,
0 = 22.2o and 0 = 15.7o . As L increases, the angle 0 tends to zero.
There are other antenna structures that act as traveling-wave antennas, as shown
in Fig. 17.6.3. For example, a waveguide with a long slit along its length will radiate
continuously along the slit. Another example is a corrugated conducting surface along
which a surface wave travels and gets radiated when it reaches the discontinuity at the
end of the structure.
In all of these examples, the radiation pattern has an angular dependence similar to
that of a linear antenna with a traveling-wave current of the form:
0zl
(17.6.7)
Fig. 17.7.1 Traveling-wave vee antenna with l = 5, 0 = 22.2o , and = 0.850 = 18.9o .
789
790
a = 0.801.00. Figure 17.7.2 shows the optimum angle factor a that corresponds to
maximum directivity (in the plane of the vee) as a function of the length l.
Optimum Angle Factor
1
0.95
0.9
0.85
Applying the result of Eq. (17.6.2), the radiation vectors of the two arms will be:
0.8
l
0.75
0
2.5
7.5
10
12.5
15
17.5
F1 =
20
l/
z1
0
l
F2 =
z2
Figure 17.7.3 shows the actual power patterns for the cases l = 5 and l = 10. The
main lobe angles were 0 = 22.2o and 0 = 15.7o . The optimum vee angles were found
to be approximately (see Fig. 17.7.2), = 0.850 = 18.9o and = 0.950 = 14.9o , in
the two cases.
L = 5, = 18.9o
0o
45o
45o
45o
I 1 ejkl(1cos 1 )
jk
1 cos 1
I 1 ejkl(1cos 2 )
jk
1 cos 2
F1 =
I
F(1 ) ,
jk
F2 =
I
F(2 )
jk
where the function F() was dened in Eq. (17.6.3). From Fig. 17.7.4, we may express
1 , 2 in terms of the polar angle with respect to the z-axis as:
L = 10, = 14.9o
0o
z1
1 = ,
45o
2 = +
90o
6 3
dB
90o
90o
6 3
dB
90o
F = F1 + F2 =
I
I
F(2 )F(1 ) =
F( + )F( )
jk
jk
135o
135o
135o
180o
180o
U()=
k2
322
|F ()|2 =
|I|2
322
2
F( + )F( )
The combined radiation pattern can be obtained with the help of Fig. 17.7.4. Let
z1 and
z2 be the two unit vectors along the two arms of the vee, and let 1 , 2 be the
two polar angles of the observation point P with respect to the directions
z1 ,
z2 . The
assumed currents along the two arms have opposite amplitudes and are:
I1 (z1 )= Iejkz1 ,
I2 (z2 )= Iejkz2 ,
for
0 z1 , z2 l
2
g()= cn F( + )F( )
(17.7.1)
This is the gain plotted in Fig. 17.7.3 and can be computed by the MATLAB function
vee. Finally, we consider briey a rhombic antenna made up of two concatenated vee
antennas, as shown in Fig. 17.7.5. Now the two inner main lobes of the rst vee (lobes
a, b) and the two outer lobes of the second vee (lobes c, d) align with each other, thus
increasing the directivity of the antenna system.
The radiation vectors F3 and F4 of arms 3 and 4 may be obtained by noting that
these arms are the translations of arms 1 and 2, and therefore, the radiation vectors are
changed by the appropriate translational phase shift factors, as discussed in Sec. 20.2.
791
792
L = 10, = 15.7o
45o
6 3
dB
45o
45o
90o
0o
90o
45o
90o
6 3
dB
90o
135o
F4 = ejkd1 F2
180o
180
For such small loops, the radiation pattern turns out to be independent of the shape
of the loop and the radiation vector takes the simple form:
F3 = ejkl cos 2 F1 =
I jkl cos 2
e
F(1 )
jk
F4 = ejkl cos 1 F2 =
135
(17.7.2)
where the negative signs arise because the currents in those arms have opposite signs
with their parallel counterparts. The phase shift factors are:
135o
135
I jkl cos 1
e
F(2 )
jk
F = jmk
where m is the loops magnetic moment dened with respect to Fig. 17.8.1 as follows:
m=
z IS ,
(magnetic moment)
F ()
F = j m k = jmk sin
I
F(2 )F(1 )+ejkl cos 2 F(1 )ejkl cos 1 F(2 )
jk
2
g()= cn F( + )F( )+ejkl cos(+) F( )ejkl cos() F( + )
Figure 17.7.6 shows the power gain g() for the cases L = 5 and L = 10. The
optimum vee angle in both cases was found to be = 0 , that is, = 22.2o and
= 15.7o . The function rhombic may be used to evaluate this expression.
(17.8.2)
sin , we have:
where S is the area of the loop. Writing k = k
r and noting that
z
r=
F = F1 + F2 + F3 + F4 =
(17.8.1)
(17.8.3)
793
jkr
4r
= mk2 sin
F
jkr
4r
sin )(z
)
z cos +
z +
k r = k(
)
= kz cos + k sin (
(17.8.4)
k2
k4 |m|2
|F |2 =
sin2
2
32
322
jkr
jkr
H = jk e
= mk2 sin e
F
H=
4r
4r
U(, )=
794
(loop intensity)
k r = ka sin cos( )
(17.8.5)
Prad
k4 |m|2 8
k4 |m|2
= Umax =
=
2
32
3
12
Replacing m by IS, we may obtain the loops radiation resistance from the denition:
Prad
1
k4 |IS|2
= Rrad |I|2 =
2
12
Rrad
k4 S2
=
6
dipole
loop
=j
Il
mk
If we choose Il = mk, then the electric elds are off by a 90 -degree phase. If
such a Hertzian dipole and loop are placed at the origin, the produced net electric eld
will be circularly polarized. We note nally that the loop may have several turns, thus
increasing its radiation resistance and radiated power. For a loop with n turns, we must
make the replacement m nm.
Next, we consider the circular loop in more detail, and derive Eq. (17.8.3). Assuming an
innitely thin wire loop of radius a, the assumed current density can be expressed in
cylindrical coordinates as in Eq. (17.1.3):
( a)(z )
J(r )= I
The radiation vector will be:
F=
jkr
J(r )e
V
3
(17.9.2)
, we have:
point P. Resolving into the directions ,
=
cos( )
sin( )
2
F = Ia
cos
sin )ejka sin cos d
(
F = Ia
2
(17.9.3)
J1 (x)=
2
1
2j
we may replace the -integral by 2jJ1 (ka sin ) and write Eq. (17.9.3) as:
Comparing Eq. (17.8.4) to the Hertzian dipole, the loops electric eld is in the direction, whereas the Hertzian dipoles is in the -direction. The relative amplitudes
of the electric elds are:
2
F = Ia
d r =
= 2j Ia J1 (ka sin )
F = F
This gives the radiation vector for any loop radius. If the loop is electrically small,
that is, ka 1, we may use the rst-order approximation J1 (x) x/2, to get
= 2j Ia
F = F
1
= jIa2 k sin
ka sin
2
(17.9.4)
(17.9.5)
795
796
The square loop of Fig. 17.8.1 may be thought of as four separate linear antennas representing the four sides. Assuming that each side is a Hertzian dipole and that the sides
are at distances l/2 from the origin, we can write the current densities of the sides
1, 2, 3, 4 as follows:
Il (x l/2)(y)(z)
J1 (r) = y
F=
Il (x + l/2)(y)(z)
J3 (r) = y
Il (x)(y + l/2)(z)
x
(x + l/2)(x l/2)
(y)(z)
l
where we multiplied and divided by a factor of l. In the limit of small l, we may replace
the quantity in the bracket by the derivative (x) of the delta function (x):
(x)(y)(z)
J1 (r)+J3 (r)= Il2 y
Similarly, we nd for sides 2 and 4:
F = Il2
Dij =
(x ) (y )y
(x )(y ) (z )ej(kx x +ky y
x
+k
zz
)
1 + j k r +
1
(jk r )2 + J(r )d3 r
2!
(17.11.1)
r (r ) d3 r
(17.11.2)
(17.11.3)
Thus, the net current density of all sides is:
p=
(17.10.1)
The rst term is the electric dipole radiation term and corresponds to the Hertzian
dipole antenna. The second term incorporates both the magnetic dipole (corresponding
to a Hertzian loop antenna) and the electric quadrupole terms.
Higher multipoles arise from the higher-order terms in the above expansion. A systematic discussion of all multipole radiation terms requires the use of spherical harmonics.
Keeping only a few terms in the above expansion is a good approximation to F provided kr 1, or l , where l is the typical dimension of the current source. In
general, any radiating system will emit radiation of various multipole types.
The electric dipole and electric quadrupole moments of a charge distribution are dened in terms of the following rst- and second-order moments of the charge density:
(x) (y)(z)
J2 (r)+J4 (r)= Il2 x
(x) (y)y
(x)(y) (z)
J(r)= Il2 x
J(r )d3 r +
elec. dipole
=
Il (x)(y l/2)(z)
J2 (r) = x
J4 (r) =
J(r )ejkr d3 r =
The identity of Problem 15.2 is useful here in manipulating the successive expansion
terms of F. Applying the identity with the two choices: g(r )= ri and g(r )= ri rj , we
obtain the relationships:
+ jkx y
)
F = Il2 (jky x
Ji d3 r = j
(ri Jj + rj Ji ) d3 r = j
V
sin + y
cos )= jIl2 k sin
F = jIl2 k sin (x
which agrees with Eq. (17.8.3), with m = IS = Il2 .
(17.10.2)
J(r ) d3 r = j p = Fel
In the second term of Eq. (17.11.1), we may apply the vectorial identity:
(k r )J =
1
1
(r J)k +
(k r )J + (k J)r ]
2
2
V
(k r )J d3 r =
1
2
V
(r J)k d3 r +
1
2
V
(k r )J + (k J)r ] d3 r
(17.11.5)
797
m=
1
2
r J(r ) d3 r
(magnetic moment)
j(k r )J d3 r = j m k
1
Dk = Fmag + Fquad
2
(17.11.7)
1
Dk
2
(17.11.8)
We briey discuss each term. For a Hertzian dipole antenna with J(r )=
z Il 3 (r ),
only the rst term of (17.11.8) is non-zero and is the same as that of Sec. 17.2:
Fel =
J(r ) d3 r =
z Il = j p
q2 a2rms
q2 |a|2
P=
=
12c2
6c2
where arms = |a|/ 2 is the rms value of the acceleration. This is Larmors classical
expression for the radiated power from a nonrelativistic accelerated charge.
For a Hertzian loop, only the magnetic moment term is present in F. We may verify
the result that m =
z IS using the denition (17.11.6). Indeed, for a circular loop:
m=
1
2
( a)(z ) d d dz
r I
.
The integrations over z and force z = 0 and = a, and therefore, r = a
=
Noting that
z and that the -integration contributes a factor of 2, we obtain:
m=
1
Ia 2 =
z I(a2 )
a
2
Similarly, inserting Eq. (17.10.1) into (17.11.6), we nd for the square loop:
m=
1
2
(17.11.6)
Therefore, the rst term in Eq. (17.11.5) may be written as m k. With the help of
the second identity of Eq. (17.11.4), the last term of (17.11.5) may be written in terms of
the quadrupole matrix D acting on the vector k. We have then for the second term in
the expansion (17.11.1):
For the electric quadrupole term, the matrix D is sometimes replaced by its traceless
version dened by
798
+yy
+ z
(x) (y)y
(x)(y) (z) dx dy dz =
z) Il2 x
z Il2
(x x
Q = 3D I tr(D)
Dk =
1
1
Q k + tr(D) k
3
3
The second term may be ignored because it does not contribute to the radiation
elds, which depend only on the part of F transverse to k. Thus, without loss of generality we may also write:
1
F = j p + j m k Q k
6
The electric and magnetic dipoles have angular gain patterns that are identical to
the Hertzian dipole and Hertzian loop antennas, that is, sin2 . The quadrupole term,
on the other hand, can have a complicated angular pattern as can be seen by expressing
the vector Q k = kQ
r explicitly in terms of the angles , :
Qxx
Q
r = Qyx
Qzx
Qxy
Qyy
Qzy
Qxz
sin cos
17.12 Problems
17.1 Computer ExperimentDipoles. Reproduce the results and graphs of Fig. 17.4.2, and calculate the corresponding directivities in dB.
17.2 Derive Eq. (17.3.7) for the input resistance of a dipole antenna.
17.3 Derive Eq. (17.6.6) for the tilt angle of a traveling wave antenna by reducing the problem to
that of nding the maximum of the function sin2 (x)/x in the interval [0, 1].
17.4 Computer ExperimentTraveling Wave Antennas. Reproduce the results and graphs of Fig. 17.6.2.